
from PIL import Image
import requests
from io import BytesIO
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import warnings
from bs4 import BeautifulSoup
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
import nltk
import math
import time
import re
import os
from tqdm import tqdm
import seaborn as sns
from collections import Counter
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity  
from sklearn.metrics import pairwise_distances
from matplotlib import gridspec
from scipy.sparse import hstack
import plotly
import plotly.figure_factory as ff
from plotly.graph_objs import Scatter, Layout
import plotly.express as px
import plotly.graph_objects as go
from pandas_profiling import ProfileReport

plotly.offline.init_notebook_mode(connected=True)
warnings.filterwarnings("ignore")


from pprint import pprint
data = pd.read_json('tops_fashion.json')
print ('Number of data points : ', data.shape[0], \
       'Number of features/variables:', data.shape[1])

pprint(data.columns)
data.to_pickle('pickels/'+str(data.shape[0])+'_women_shirts_data')

Number of data points :  183138 Number of features/variables: 19
Index(['sku', 'asin', 'product_type_name', 'formatted_price', 'author',
       'color', 'brand', 'publisher', 'availability', 'reviews',
       'large_image_url', 'availability_type', 'small_image_url',
       'editorial_review', 'title', 'model', 'medium_image_url',
       'manufacturer', 'editorial_reivew'],
      dtype='object')


data=data[data['product_type_name']=='SHIRT']


data = data[['asin', 'brand', 'color', 'medium_image_url', 'product_type_name', 'title', 'formatted_price']]
print ('Number of data points : ', data.shape[0], \
       'Number of features:', data.shape[1])
data.head()

Number of data points :  167794 Number of features: 7


data = data.loc[~data['formatted_price'].isnull()]
print('Number of data points After eliminating price=NULL :', data.shape[0])

data =data.loc[~data['color'].isnull()]
print('Number of data points After eliminating color=NULL :', data.shape[0])

Number of data points After eliminating price=NULL : 21513
Number of data points After eliminating color=NULL : 21504


mkdir Raw_images\\SHIRTS

A subdirectory or file Raw_images\\SHIRTS already exists.


'''
url_not_found =[]
index_not_found =[]
for index,row in tqdm(data.iterrows()):
        url = row['medium_image_url']
        response = requests.get(url)
        try:
            img = Image.open(BytesIO(response.content))
            img.save('Raw_images/SHIRTS/'+row['asin']+'.jpeg')
            
        except:
            url_not_found.append(url)
            index_not_found.append(index)'''

"#for dowloading images from the json file containg the links of the images \nurl_not_found =[]\nindex_not_found =[]\nfor index,row in tqdm(data.iterrows()):\n        url = row['medium_image_url']\n        response = requests.get(url)\n        try:\n            img = Image.open(BytesIO(response.content))\n            img.save('Raw_images/SHIRTS/'+row['asin']+'.jpeg')\n            \n        except:\n            url_not_found.append(url)\n            index_not_found.append(index)"


files_downloaded=os.listdir('Raw_images/SHIRTS/')

for i in range(0,len(files_downloaded)):
    
    files_downloaded[i]=files_downloaded[i].replace('.jpeg','')


data = data[data.asin.isin(files_downloaded)]


data.reset_index(drop=True,inplace=True)
data


profile = ProfileReport(data, title="Amazon Women Shirts Data Report")

profile.to_notebook_iframe()

Summarize dataset:   0%|          | 0/5 [00:00<?, ?it/s]

Generate report structure:   0%|          | 0/1 [00:00<?, ?it/s]

Render HTML:   0%|          | 0/1 [00:00<?, ?it/s]


print('Number of data points After eliminating images which were not accesible :', data.shape[0])

Number of data points After eliminating images which were not accesible : 14338


data.to_pickle('pickels/'+str(data.shape[0])+'_women_shirts_data')


data_sorted = data[data['title'].apply(lambda x: len(x.split())>4)]
print("After removal of products with short description:", data_sorted.shape[0])

After removal of products with short description: 14136


print('The number of entries with duplicate title is %d'%sum(data.duplicated('title')))
data_sorted.sort_values('title',inplace=True, ascending=False)

indices = []
for i,row in data_sorted.iterrows():
    indices.append(i)
import itertools
stage1_dedupe_asins = []
i = 0
j = 0
num_data_points = data_sorted.shape[0]
while i < num_data_points and j < num_data_points:    
    previous_i = i
    a = data['title'].loc[indices[i]].split()
    
    j = i+1
    while j < num_data_points:
        
        b = data['title'].loc[indices[j]].split()
        
        length = max(len(a), len(b))
        
        count  = 0
        
        
        
        
        for k in itertools.zip_longest(a,b): 
            if (k[0] == k[1]):
                count += 1
        
        
        if (length - count) > 2: 
            
            stage1_dedupe_asins.append(data_sorted['asin'].loc[indices[i]])
            
            if j == num_data_points-1: stage1_dedupe_asins.append(data_sorted['asin'].loc[indices[j]])
            
            i = j
            break
        else:
            j += 1
    if previous_i == i:
        break

data = data.loc[data['asin'].isin(stage1_dedupe_asins)]
print('Number of data points now is: ', data.shape[0])
data.to_pickle('pickels/'+str(data.shape[0])+'_women_shirts_data')

The number of entries with duplicate title is 839
Number of data points now is:  9911


%%time


indices = []
for i,row in data.iterrows():
    indices.append(i)

stage2_dedupe_asins = []
while len(indices)!=0:
    i = indices.pop()
    stage2_dedupe_asins.append(data['asin'].loc[i])
    
    a = data['title'].loc[i].split()
    
    for j in indices:        
        b = data['title'].loc[j].split()
        
        length = max(len(a),len(b))        
        
        count  = 0
        
        
        
        
        for k in itertools.zip_longest(a,b): 
            if (k[0]==k[1]):
                count += 1
        
        if (length - count) < 3:
            indices.remove(j)
            
data = data.loc[data['asin'].isin(stage2_dedupe_asins)]
print('Number of data points after stage two of dedupe: ',data.shape[0])
data.to_pickle('pickels/'+str(data.shape[0])+'_women_shirts_data')

Number of data points after stage two of dedupe:  9391
Wall time: 17min 40s


stop_words = set(stopwords.words('english'))
print ('list of stop words:', stop_words)

def nlp_preprocessing(total_text, index, column):
    if type(total_text) is not int:
        string = ""
        for words in total_text.split():
            
            word = ("".join(e for e in words if e.isalnum()))
            
            word = word.lower()
            
            if not word in stop_words:
                string += word + " "
        data[column][index] = string

list of stop words: {'she', 'couldn', 'did', 'you', "that'll", 'we', "you're", "shouldn't", 'up', 'his', 'just', 'above', 'was', "couldn't", 'here', 'me', 'its', 'then', 'y', 'few', 'these', 'he', 'own', 'because', 'isn', 'when', "it's", 'of', 'no', 'with', 'will', 're', 'ma', 'wasn', 'their', 'the', 'hasn', "she's", 'do', 't', 'if', 'from', 'doesn', "haven't", 'this', "you've", 'once', 'who', 'your', 'yourself', "hadn't", 'there', "you'll", 'i', 'our', 'are', 'most', 'd', 'at', 'during', 'mustn', 'had', 'under', 'themselves', 'other', 've', "isn't", 'about', 'nor', 'any', 'shan', 'they', 'further', 'ain', "didn't", "doesn't", 'now', "you'd", 'for', 'both', 'too', 'having', 'itself', 'before', 'each', "wasn't", 'my', 'can', 'shouldn', 'until', 'be', 'were', 'some', 'hers', 'him', 'aren', 'didn', "weren't", "wouldn't", 'only', 'through', 'as', 's', 'weren', 'which', 'very', 'herself', 'what', 'but', 'to', 'himself', "aren't", 'so', 'needn', 'all', 'in', 'down', 'or', 'whom', "don't", 'wouldn', 'not', 'off', 'than', 'them', 'those', 'same', 'after', 'o', 'have', 'been', 'between', 'ourselves', 'such', "mightn't", "shan't", "needn't", 'why', "mustn't", 'on', 'out', 'how', 'where', 'is', 'm', 'a', 'haven', 'into', 'her', 'while', 'ours', 'won', 'am', 'myself', 'yourselves', 'theirs', "hasn't", 'an', 'don', "won't", 'by', 'more', 'hadn', "should've", 'll', 'that', 'against', 'it', 'again', 'yours', 'below', 'mightn', 'doing', 'has', 'being', 'does', 'and', 'should', 'over'}


start_time = time.time()

for index, row in data.iterrows():
    nlp_preprocessing(row['title'], index, 'title')

print(time.time() - start_time, "seconds")

data.head()
data.to_pickle('pickels/'+str(data.shape[0])+'_women_shirts_data')

3.0838680267333984 seconds


def display_img(url,ax,fig):
    
    response = requests.get(url)
    img = Image.open(BytesIO(response.content))
    
    plt.imshow(img)


def plot_heatmap(keys, values, labels, url, text):
        gs = gridspec.GridSpec(2, 2, width_ratios=[4,1], height_ratios=[4,1]) 
        fig = plt.figure(figsize=(25,3))
        
        
        ax = plt.subplot(gs[0])
        
        ax = sns.heatmap(np.array([values]), annot=np.array([labels]))
        ax.set_xticklabels(keys) 
        ax.set_title(text) 
        
        
        ax = plt.subplot(gs[1])
        
        ax.grid(False)
        ax.set_xticks([])
        ax.set_yticks([])
        
        
        display_img(url, ax, fig)
        
        
        plt.show()
    
def plot_heatmap_image(doc_id, vec1, vec2, url, text, model):
    
    intersection = set(vec1.keys()) & set(vec2.keys()) 
    
    for i in vec2:
        if i not in intersection:
            vec2[i]=0
    
    keys = list(vec2.keys())
    
    values = [vec2[x] for x in vec2.keys()]    
    
        
        
        
    if model == 'bag_of_words':
        labels = values
    elif model == 'tfidf':
        labels = []
        for x in vec2.keys():
            
            
            if x in  tfidf_title_vectorizer.vocabulary_:
                labels.append(tfidf_title_features[doc_id, tfidf_title_vectorizer.vocabulary_[x]])
            else:
                labels.append(0)
    elif model == 'idf':
        labels = []
        for x in vec2.keys():
            
            
            if x in  idf_title_vectorizer.vocabulary_:
                labels.append(idf_title_features[doc_id, idf_title_vectorizer.vocabulary_[x]])
            else:
                labels.append(0)
    plot_heatmap(keys, values, labels, url, text)


def text_to_vector(text):
    word = re.compile(r'\w+')
    words = word.findall(text)
    
    
    return Counter(words) 
    

def get_result(doc_id, content_a, content_b, url, model):
    text1 = content_a
    text2 = content_b    
    
    vector1 = text_to_vector(text1)
    
    vector2 = text_to_vector(text2)
    plot_heatmap_image(doc_id, vector1, vector2, url, text2, model)


from sklearn.feature_extraction.text import CountVectorizer
title_vectorizer = CountVectorizer()
title_features   = title_vectorizer.fit_transform(data['title'])
title_features.get_shape() 



bag_of_words_euclidean=[]
def bag_of_words_model(doc_id, num_results):
    pairwise_dist = pairwise_distances(title_features,title_features[doc_id])    
    indices = np.argsort(pairwise_dist.flatten())[0:num_results]
    
    pdists  = np.sort(pairwise_dist.flatten())[0:num_results]

    
    df_indices = list(data.index[indices])
    
    for i in range(0,len(indices)):
        
        get_result(indices[i],data['title'].loc[df_indices[0]], data['title'].loc[df_indices[i]], data['medium_image_url'].loc[df_indices[i]], 'bag_of_words')
        print('ASIN :',data['asin'].loc[df_indices[i]])
        print ('Brand:', data['brand'].loc[df_indices[i]])
        print ('Title:', data['title'].loc[df_indices[i]])
        bag_of_words_euclidean.append(pdists[i])
        print ('Euclidean similarity with the query image :', pdists[i])
        print('='*60)
    print('Average euclidean distance is ',sum(bag_of_words_euclidean)/num_results)


print('Getting the similar items for document id and number of items')
bag_of_words_model(998, 15)

Getting the similar items for document id and number of items

ASIN : B00VJ19SSO
Brand: Active Products
Title: 2 pack active basic womens shelf bra cami top small black black 
Euclidean similarity with the query image : 0.0
============================================================

ASIN : B00K5BHL06
Brand: Active USA
Title: 2 pack active basic womens long cami w built bra small black black 
Euclidean similarity with the query image : 2.0
============================================================

ASIN : B00VIU79DW
Brand: Active Products
Title: 2 pack active basic womens basic cami tank top small black white 
Euclidean similarity with the query image : 2.449489742783178
============================================================

ASIN : B00HDZM2RO
Brand: Zenana Outfitters
Title: 2 pack zenana womens basic ribbed tank top small black red 
Euclidean similarity with the query image : 3.0
============================================================

ASIN : B075831F14
Brand: Very J
Title: j womens tie dye top black small 
Euclidean similarity with the query image : 3.0
============================================================

ASIN : B07258X2FK
Brand: Who What Wear
Title: wear womens ruffle top small black 
Euclidean similarity with the query image : 3.0
============================================================

ASIN : B01N44FN88
Brand: Project Pinup
Title: lets mermaids womens black cami tank top small 
Euclidean similarity with the query image : 3.0
============================================================

ASIN : B01MXOSQR4
Brand: Project Pinup
Title: strange  unusual womens black cami tank top small 
Euclidean similarity with the query image : 3.0
============================================================

ASIN : B01NCB6I63
Brand: Project Pinup
Title: stay weird womens black cami tank top small 
Euclidean similarity with the query image : 3.0
============================================================


tfidf_title_vectorizer = TfidfVectorizer(min_df = 0)
tfidf_title_features = tfidf_title_vectorizer.fit_transform(data['title'])

tf_idf_euclidean=[]
def tfidf_model(doc_id, num_results):
    pairwise_dist = pairwise_distances(tfidf_title_features,tfidf_title_features[doc_id])

    indices = np.argsort(pairwise_dist.flatten())[0:num_results]

    pdists  = np.sort(pairwise_dist.flatten())[0:num_results]

    df_indices = list(data.index[indices])

    for i in range(0,len(indices)):
        
        get_result(indices[i], data['title'].loc[df_indices[0]], data['title'].loc[df_indices[i]],data['medium_image_url'].loc[df_indices[i]], 'tfidf')
        print('ASIN :',data['asin'].loc[df_indices[i]])
        print('BRAND :',data['brand'].loc[df_indices[i]])
        tf_idf_euclidean.append(pdists[i])
        print ('Eucliden distance from the given image :', pdists[i])
        print('='*125)
    print('Average euclidean distance is',sum(tf_idf_euclidean)/num_results)


print('Getting the similar items for document id and number of items')
tfidf_model(66, 15)

Getting the similar items for document id and number of items

ASIN : B0148RPYQK
BRAND : Five Finger Death Punch
Eucliden distance from the given image : 0.0
=============================================================================================================================

ASIN : B00Y125DYQ
BRAND : Bay Island Sportswear
Eucliden distance from the given image : 1.2165614771893531
=============================================================================================================================

ASIN : B00WA9MXLI
BRAND : Queen
Eucliden distance from the given image : 1.2182232235584929
=============================================================================================================================

ASIN : B0751JGSCS
BRAND : Premise Studio
Eucliden distance from the given image : 1.2555310786185432
=============================================================================================================================

ASIN : B071YSN9YY
BRAND : Mossimo Supply Co
Eucliden distance from the given image : 1.2703387736934366
=============================================================================================================================

ASIN : B06WVLSG56
BRAND : PINK ROSE
Eucliden distance from the given image : 1.2747444052507093
=============================================================================================================================

ASIN : B015H3W0TS
BRAND : KINGDE
Eucliden distance from the given image : 1.2792767914595835
=============================================================================================================================

ASIN : B00X6PG41A
BRAND : MKP Crop Top
Eucliden distance from the given image : 1.279342200913531
=============================================================================================================================

ASIN : B074H47ZBD
BRAND : Joyce Clapham
Eucliden distance from the given image : 1.2808232647379771
=============================================================================================================================


idf_title_vectorizer = CountVectorizer()
idf_title_features = idf_title_vectorizer.fit_transform(data['title'])


def n_containing(word):
    return sum(1 for blob in data['title'] if word in blob.split())

def idf(word):
    return math.log(data.shape[0] / (n_containing(word)))


%%time
idf_title_features  = idf_title_features.astype(np.float)

for i in idf_title_vectorizer.vocabulary_.keys():
    idf_val = idf(i)
    
    for j in idf_title_features[:, idf_title_vectorizer.vocabulary_[i]].nonzero()[0]:    
        idf_title_features[j,idf_title_vectorizer.vocabulary_[i]] = idf_val

Wall time: 1min 51s


idf_euclidean=[]
def idf_model(doc_id, num_results):
    pairwise_dist = pairwise_distances(idf_title_features,idf_title_features[doc_id])
    indices = np.argsort(pairwise_dist.flatten())[0:num_results]
    pdists  = np.sort(pairwise_dist.flatten())[0:num_results]
    df_indices = list(data.index[indices])
    for i in range(0,len(indices)):
        get_result(indices[i],data['title'].loc[df_indices[0]], data['title'].loc[df_indices[i]], data['medium_image_url'].loc[df_indices[i]], 'idf')
        print('ASIN :',data['asin'].loc[df_indices[i]])
        print('Brand :',data['brand'].loc[df_indices[i]])
        idf_euclidean.append(pdists[i])
        print ('euclidean distance from the given image :', pdists[i])
        print('='*125)
    print('Average euclidean distance is ',sum(idf_euclidean)/num_results)


idf_model(1466, 15)

ASIN : B06XXS7683
Brand : Alexis Taylor
euclidean distance from the given image : 0.0
=============================================================================================================================

ASIN : B074T9KG9Q
Brand : Rain
euclidean distance from the given image : 17.006762005047516
=============================================================================================================================

ASIN : B00KF2N5PU
Brand : Vietsbay
euclidean distance from the given image : 17.191284126497184
=============================================================================================================================

ASIN : B07581F4DG
Brand : Very J
euclidean distance from the given image : 17.307548928160667
=============================================================================================================================

ASIN : B07567M2M9
Brand : Sunshine
euclidean distance from the given image : 17.38409424086828
=============================================================================================================================

ASIN : B0758WDS6H
Brand : Pleione
euclidean distance from the given image : 17.462601457626473
=============================================================================================================================

ASIN : B074G5G5RK
Brand : ERMANNO SCERVINO
euclidean distance from the given image : 17.505125169803332
=============================================================================================================================

ASIN : B071ZSHMPC
Brand : H By Bordeaux
euclidean distance from the given image : 17.506212466863413
=============================================================================================================================

ASIN : B06XWYL8XD
Brand : BCX
euclidean distance from the given image : 17.527596182043442
=============================================================================================================================

ASIN : B00JMA25AW
Brand : Girltalk clothing
euclidean distance from the given image : 17.529903054592896
=============================================================================================================================


'''import gensim.downloader as api
word2vec_model = api.load('word2vec-google-news-300')
import pickle

file = open('word2vec_model.pkl', 'wb')
pickle.dump(word2vec_model, file)
file.close()'''

"import gensim.downloader as api\nword2vec_model = api.load('word2vec-google-news-300')\nimport pickle\n\nfile = open('word2vec_model.pkl', 'wb')\npickle.dump(word2vec_model, file)\nfile.close()"


from gensim.models import Word2Vec
from gensim.models import KeyedVectors
import pickle
with open('word2vec_model.pkl', 'rb') as handle:
    model = pickle.load(handle)


def get_word_vec(sentence, doc_id, m_name):
    vec = []
    for i in sentence.split():
        if i in vocab:
            if m_name == 'weighted' and i in  idf_title_vectorizer.vocabulary_:
                vec.append(idf_title_features[doc_id, idf_title_vectorizer.vocabulary_[i]] * model[i])
            elif m_name == 'avg':
                vec.append(model[i])
        else:

            vec.append(np.zeros(shape=(300,)))

    return  np.array(vec)

def get_distance(vec1, vec2):

    
    final_dist = []
    for i in vec1:
        dist = []
        for j in vec2:
            dist.append(np.linalg.norm(i-j))
        final_dist.append(np.array(dist))
    return np.array(final_dist)


def heat_map_w2v(sentence1, sentence2, url, doc_id1, doc_id2, model):

    s1_vec = get_word_vec(sentence1, doc_id1, model)
    
    s2_vec = get_word_vec(sentence2, doc_id2, model)
    
    
    s1_s2_dist = get_distance(s1_vec, s2_vec)

    gs = gridspec.GridSpec(2, 2, width_ratios=[4,1],height_ratios=[2,1]) 
    fig = plt.figure(figsize=(15,15))
    
    ax = plt.subplot(gs[0])
    
    ax = sns.heatmap(np.round(s1_s2_dist,4), annot=True)
    
    ax.set_xticklabels(sentence2.split())
    
    ax.set_yticklabels(sentence1.split())
    
    ax.set_title(sentence2)
    
    ax = plt.subplot(gs[1])
    
    ax.grid(False)
    ax.set_xticks([])
    ax.set_yticks([])
    display_img(url, ax, fig)
    
    plt.show()


vocab = model.index_to_key 
def build_avg_vec(sentence, num_features, doc_id, m_name):
    featureVec = np.zeros((num_features,), dtype="float32")
    nwords = 0
    for word in sentence.split():
        nwords += 1
        if word in vocab:
            if m_name == 'weighted' and word in  idf_title_vectorizer.vocabulary_:
                featureVec = np.add(featureVec, idf_title_features[doc_id, idf_title_vectorizer.vocabulary_[word]] * model[word])
            elif m_name == 'avg':
                featureVec = np.add(featureVec, model[word])
    if(nwords>0):
        featureVec = np.divide(featureVec, nwords)
    
    return featureVec


%%time
doc_id = 0
w2v_title = []
for i in data['title']:
    w2v_title.append(build_avg_vec(i, 300, doc_id,'avg'))
    doc_id += 1
w2v_title = np.array(w2v_title)

Wall time: 18min 5s


avg_w2v_euclidean=[]
def avg_w2v_model(doc_id, num_results):
        
    pairwise_dist = pairwise_distances(w2v_title, w2v_title[doc_id].reshape(1,-1))
    
    indices = np.argsort(pairwise_dist.flatten())[0:num_results]
    
    pdists  = np.sort(pairwise_dist.flatten())[0:num_results]

    
    df_indices = list(data.index[indices])
    
    for i in range(0, len(indices)):
        heat_map_w2v(data['title'].loc[df_indices[0]],data['title'].loc[df_indices[i]], data['medium_image_url'].loc[df_indices[i]], indices[0], indices[i], 'avg')
        print('ASIN :',data['asin'].loc[df_indices[i]])
        print('BRAND :',data['brand'].loc[df_indices[i]])
        avg_w2v_euclidean.append(pdists[i])
        print ('euclidean distance from given input image :', pdists[i])
        print('='*125)
    print('Average euclidean distance is ',sum(avg_w2v_euclidean)/num_results)


avg_w2v_model(566, 20)

ASIN : B0746ZVMJ7
BRAND : Banana Republic Factory Store
euclidean distance from given input image : 0.0
=============================================================================================================================

ASIN : B0758PTMYJ
BRAND : Melrose & Market
euclidean distance from given input image : 0.81187904
=============================================================================================================================

ASIN : B00LQD3J8A
BRAND : L.K. Bennett
euclidean distance from given input image : 0.81762654
=============================================================================================================================

ASIN : B06XNXRK6K
BRAND : Banana Republic
euclidean distance from given input image : 0.84567016
=============================================================================================================================

ASIN : B00OTB3KJO
BRAND : Miraclebody by Miraclesuit
euclidean distance from given input image : 0.8591212
=============================================================================================================================

ASIN : B01I7DCWFC
BRAND : KIIND OF
euclidean distance from given input image : 0.8738429
=============================================================================================================================

ASIN : B0758VCXC8
BRAND : Daniel Rainn
euclidean distance from given input image : 0.8769885
=============================================================================================================================

ASIN : B06XZSQ467
BRAND : Banana Republic
euclidean distance from given input image : 0.8774647
=============================================================================================================================

ASIN : B06XX33JLB
BRAND : French Laundry
euclidean distance from given input image : 0.8875258
=============================================================================================================================

ASIN : B00DQ4ZDWC
BRAND : W118 by Walter Baker
euclidean distance from given input image : 0.89426047
=============================================================================================================================


%%time
doc_id = 0
w2v_title_weight = []

for i in data['title']:
    w2v_title_weight.append(build_avg_vec(i, 300, doc_id,'weighted'))
    doc_id += 1

w2v_title_weight = np.array(w2v_title_weight)

Wall time: 18min 8s


weighted_w2v_euclidean=[]
def weighted_w2v_model(doc_id, num_results):    
    pairwise_dist = pairwise_distances(w2v_title_weight, w2v_title_weight[doc_id].reshape(1,-1))
    indices = np.argsort(pairwise_dist.flatten())[0:num_results]
    pdists  = np.sort(pairwise_dist.flatten())[0:num_results]
    df_indices = list(data.index[indices])
    try:
        for i in range(0, len(indices)):
            heat_map_w2v(data['title'].loc[df_indices[0]],data['title'].loc[df_indices[i]], data['medium_image_url'].loc[df_indices[i]], indices[0], indices[i], 'weighted')
            print('ASIN :',data['asin'].loc[df_indices[i]])
            print('Brand :',data['brand'].loc[df_indices[i]])
            weighted_w2v_euclidean.append(pdists[i])
            print('euclidean distance from input :', pdists[i])
            print('='*125)
    except:
        pass
    print('Average euclidean distance is ',sum(weighted_w2v_euclidean)/num_results)


weighted_w2v_model(126, 10)

ASIN : B06XTY2F8P
Brand : Feel The Piece
euclidean distance from input : 0.0
=============================================================================================================================

ASIN : B071X149W9
Brand : Feel The Piece
euclidean distance from input : 3.1869981
=============================================================================================================================

ASIN : B074VXYKZK
Brand : Feel The Piece
euclidean distance from input : 3.4808264
=============================================================================================================================

ASIN : B0759X6NV8
Brand : Serena Williams
euclidean distance from input : 4.5262566
=============================================================================================================================
Average euclidean distance is  1.1194081068038941


data['brand'].fillna(value="Not given", inplace=True )
brands = [x.replace(" ", "-") for x in data['brand'].values]
types = [x.replace(" ", "-") for x in data['product_type_name'].values]
colors = [x.replace(" ", "-") for x in data['color'].values]

brand_vectorizer = CountVectorizer()
brand_features = brand_vectorizer.fit_transform(brands)

type_vectorizer = CountVectorizer()
type_features = type_vectorizer.fit_transform(types)

color_vectorizer = CountVectorizer()
color_features = color_vectorizer.fit_transform(colors)

extra_features = hstack((brand_features, type_features, color_features)).tocsr()


def heat_map_w2v_brand(sentance1, sentance2, url, doc_id1, doc_id2, df_id1, df_id2, model):    
    s1_vec = get_word_vec(sentance1, doc_id1, model)
    
    s2_vec = get_word_vec(sentance2, doc_id2, model)
    
    
    
    s1_s2_dist = get_distance(s1_vec, s2_vec)
   
    data_matrix = [['Asin','Brand', 'Color', 'Product type'],
               [data['asin'].loc[df_id1],brands[doc_id1], colors[doc_id1], types[doc_id1]], 
               [data['asin'].loc[df_id2],brands[doc_id2], colors[doc_id2], types[doc_id2]]] 
    
    colorscale = [[0, '#1d004d'],[.5, '#f2e5ff'],[1, '#f2e5d1']]
    table = ff.create_table(data_matrix, index=True, colorscale=colorscale)
    plotly.offline.iplot(table, filename='simple_table')
    gs = gridspec.GridSpec(25, 15)
    fig = plt.figure(figsize=(25,5))
    ax1 = plt.subplot(gs[:, :-5])
    
    ax1 = sns.heatmap(np.round(s1_s2_dist,6), annot=True)
    
    ax1.set_xticklabels(sentance2.split())
    
    ax1.set_yticklabels(sentance1.split())
    
    ax1.set_title(sentance2)

    
    ax2 = plt.subplot(gs[:, 10:16])
    
    ax2.grid(False)
    ax2.set_xticks([])
    ax2.set_yticks([])
    
    
    display_img(url, ax2, fig)
    
    plt.show()


brand_colour_euclidean=[]
def idf_w2v_brand(doc_id, w1, w2, num_results):
    idf_w2v_dist  = pairwise_distances(w2v_title_weight, w2v_title_weight[doc_id].reshape(1,-1))
    ex_feat_dist = pairwise_distances(extra_features, extra_features[doc_id])
    pairwise_dist   = (w1 * idf_w2v_dist +  w2 * ex_feat_dist)/float(w1 + w2)
    indices = np.argsort(pairwise_dist.flatten())[0:num_results]
    pdists  = np.sort(pairwise_dist.flatten())[0:num_results]
    df_indices = list(data.index[indices])
    
    for i in range(0, len(indices)):
        heat_map_w2v_brand(data['title'].loc[df_indices[0]],data['title'].loc[df_indices[i]], data['medium_image_url'].loc[df_indices[i]], indices[0], indices[i],df_indices[0], df_indices[i], 'weighted')
        print('ASIN :',data['asin'].loc[df_indices[i]])
        print('Brand :',data['brand'].loc[df_indices[i]])
        brand_colour_euclidean.append(pdists[i])
        print('euclidean distance from input :', pdists[i])
        print('='*125)
    print('Average euclidean distance is ',sum(brand_colour_euclidean)/num_results)


idf_w2v_brand(490, 5, 5, 15)

ASIN : B00YONC7AU
Brand : AutumnFall
euclidean distance from input : 4.2146848500124176e-08
=============================================================================================================================

ASIN : B00YE0USAY
Brand : bestpriceam
euclidean distance from input : 2.135294532956567
=============================================================================================================================

ASIN : B00YBYYPWK
Brand : WenHong
euclidean distance from input : 2.214030075073242
=============================================================================================================================

ASIN : B010AVGFIY
Brand : MFrannie
euclidean distance from input : 2.236142921628442
=============================================================================================================================

ASIN : B06XFVRWWL
Brand : Tosangn_Clothes
euclidean distance from input : 2.2363612176794674
=============================================================================================================================

ASIN : B00QGEJ3MA
Brand : Towallmark
euclidean distance from input : 2.2571954728979735
=============================================================================================================================

ASIN : B00JKCQZJE
Brand : Demarkt
euclidean distance from input : 2.276511287869897
=============================================================================================================================

ASIN : B01A6IZM6C
Brand : Anermy
euclidean distance from input : 2.2781511308569575
=============================================================================================================================

ASIN : B01JS2A3BG
Brand : Dresswell
euclidean distance from input : 2.297120189847436
=============================================================================================================================

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
~\AppData\Local\Temp/ipykernel_11744/2920441448.py in <module>
----> 1 idf_w2v_brand(490, 5, 5, 15)

~\AppData\Local\Temp/ipykernel_11744/256987617.py in idf_w2v_brand(doc_id, w1, w2, num_results)
     21 
     22     for i in range(0, len(indices)):
---> 23         heat_map_w2v_brand(data['title'].loc[df_indices[0]],data['title'].loc[df_indices[i]], data['medium_image_url'].loc[df_indices[i]], indices[0], indices[i],df_indices[0], df_indices[i], 'weighted')
     24         print('ASIN :',data['asin'].loc[df_indices[i]])
     25         print('Brand :',data['brand'].loc[df_indices[i]])

~\AppData\Local\Temp/ipykernel_11744/1651732824.py in heat_map_w2v_brand(sentance1, sentance2, url, doc_id1, doc_id2, df_id1, df_id2, model)
     39     ax1 = sns.heatmap(np.round(s1_s2_dist,6), annot=True)
     40     # set the x axis labels as recommended apparels title
---> 41     ax1.set_xticklabels(sentance2.split())
     42     # set the y axis labels as input apparels title
     43     ax1.set_yticklabels(sentance1.split())

~\Anaconda3\envs\condaenv\lib\site-packages\matplotlib\axes\_base.py in wrapper(self, *args, **kwargs)
     61 
     62         def wrapper(self, *args, **kwargs):
---> 63             return get_method(self)(*args, **kwargs)
     64 
     65         wrapper.__module__ = owner.__module__

~\Anaconda3\envs\condaenv\lib\site-packages\matplotlib\cbook\deprecation.py in wrapper(*args, **kwargs)
    449                 "parameter will become keyword-only %(removal)s.",
    450                 name=name, obj_type=f"parameter of {func.__name__}()")
--> 451         return func(*args, **kwargs)
    452 
    453     return wrapper

~\Anaconda3\envs\condaenv\lib\site-packages\matplotlib\axis.py in _set_ticklabels(self, labels, fontdict, minor, **kwargs)
   1791         if fontdict is not None:
   1792             kwargs.update(fontdict)
-> 1793         return self.set_ticklabels(labels, minor=minor, **kwargs)
   1794 
   1795     @cbook._make_keyword_only("3.2", "minor")

~\Anaconda3\envs\condaenv\lib\site-packages\matplotlib\axis.py in set_ticklabels(self, ticklabels, minor, **kwargs)
   1713             if len(locator.locs) != len(ticklabels) and len(ticklabels) != 0:
   1714                 raise ValueError(
-> 1715                     "The number of FixedLocator locations"
   1716                     f" ({len(locator.locs)}), usually from a call to"
   1717                     " set_ticks, does not match"

ValueError: The number of FixedLocator locations (12), usually from a call to set_ticks, does not match the number of ticklabels (13).


import numpy as np
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Dropout, Flatten, Dense
from keras import applications
from sklearn.metrics import pairwise_distances
import matplotlib.pyplot as plt
import requests
from PIL import Image
import pandas as pd
import pickle
from tensorflow.keras.applications.vgg16 import VGG16

ls


%%time

img_width, img_height = 224, 224

top_model_weights_path = 'bottleneck_fc_model.h5'
train_data_dir = r'Raw_images/'
nb_train_samples = 9000
epochs = 50
batch_size = 1


def save_bottlebeck_features():
    
    asins = []
    datagen = ImageDataGenerator(rescale=1. / 255)
    
    
    model = VGG16(include_top=False, weights='imagenet')
    print('Image Net Pretrained VGG16 weights loaded')
    generator = datagen.flow_from_directory(
        train_data_dir,
        target_size=(img_width, img_height),
        batch_size=batch_size,
        class_mode=None,
        shuffle=False)
    print('All the images loaded from current directory')
    for i in generator.filenames:
        
        asins.append(i[7:-5])

    bottleneck_features_train = model.predict_generator(generator, nb_train_samples // batch_size)
    bottleneck_features_train = bottleneck_features_train.reshape((9000,25088))
    
    np.save(open('9k_data_cnn_features.npy', 'wb'), bottleneck_features_train)
    np.save(open('9k_data_cnn_feature_asins.npy', 'wb'), np.array(asins))
    

save_bottlebeck_features()

Image Net Pretrained VGG16 weights loaded
Found 14338 images belonging to 1 classes.
All the images loaded from current directory
Wall time: 16min 39s


bottleneck_features_train = np.load('9k_data_cnn_features.npy')
asins = np.load('9k_data_cnn_feature_asins.npy')
asins = list(asins)


data = pd.read_pickle('pickels/9391_women_shirts_data')
df_asins = list(data['asin'])


from IPython.display import display, Image, SVG, Math, YouTubeVideo

cnn_euclidean=[]

def get_similar_products_cnn(doc_id, num_results):
    doc_id = asins.index(df_asins[doc_id])
    pairwise_dist = pairwise_distances(bottleneck_features_train, bottleneck_features_train[doc_id].reshape(1,-1))

    indices = np.argsort(pairwise_dist.flatten())[0:num_results]
    pdists  = np.sort(pairwise_dist.flatten())[0:num_results]

    for i in range(len(indices)):
        rows = data[['medium_image_url','title']].loc[data['asin']==asins[indices[i]]]
        for indx, row in rows.iterrows():
            display(Image(url=row['medium_image_url'], embed=True))
            print('Product Title: ', row['title'])
            cnn_euclidean.append(pdists[i])
            print('Euclidean Distance from input image:', pdists[i])
            print('Amazon Url: www.amzon.com/dp/'+ asins[indices[i]])
    print('Average euclidean distance is ',sum(cnn_euclidean)/num_results)


get_similar_products_cnn(129,10)

Product Title:  fever womens sleeveless blouse red coral large 
Euclidean Distance from input image: 5.3947965e-06
Amazon Url: www.amzon.com/dp/B06XQ4ZCJK

Product Title:  hunter bell womens toppin blouse green multi extra small 
Euclidean Distance from input image: 38.171734
Amazon Url: www.amzon.com/dp/B01N1P0WH5

Product Title:  2016 tropical tank crop tops watermelon print women 
Euclidean Distance from input image: 39.063503
Amazon Url: www.amzon.com/dp/B01HUSS62I

Product Title:  drew womens tweed knit audie top sz ivorygrey 230135f 
Euclidean Distance from input image: 39.88518
Amazon Url: www.amzon.com/dp/B01ETLYQ6E

Product Title:  fapizi womens short sleeved jumpsuit tops blouse tshirt xl 
Euclidean Distance from input image: 39.952923
Amazon Url: www.amzon.com/dp/B01IT5CR7Q

Product Title:  kingde coconut tree digital printing sleeveless vestbqn29 
Euclidean Distance from input image: 40.689045
Amazon Url: www.amzon.com/dp/B015H3WPCK

Product Title:  bky v neck floral rayon blouse comfortable white bkyo129 
Euclidean Distance from input image: 40.810284
Amazon Url: www.amzon.com/dp/B00Y8FVYH6

Product Title:  women floral print chiffon tshirt good quality bky5036 
Euclidean Distance from input image: 41.79631
Amazon Url: www.amzon.com/dp/B00R1FN9AK

Product Title:  jerry leigh apparel juniors ruffle anchor graphic tank blue small 3 
Euclidean Distance from input image: 42.199135
Amazon Url: www.amzon.com/dp/B01F2UDHGG
Average euclidean distance is  32.256812602396806


euclidean_distance=[]
num_results=20
euclidean_distance.append(sum(bag_of_words_euclidean)/num_results)
euclidean_distance.append(sum(tf_idf_euclidean)/num_results)
euclidean_distance.append(sum(idf_euclidean)/num_results)
euclidean_distance.append(sum(avg_w2v_euclidean)/num_results)
euclidean_distance.append(sum(weighted_w2v_euclidean)/num_results)
euclidean_distance.append(sum(brand_colour_euclidean)/num_results)
euclidean_distance.append(sum(cnn_euclidean)/num_results)
x=euclidean_distance
y=[]
for i in range(0,47,7):
    y.append(i)


fig = go.Figure()
fig.add_trace(go.Scatter(x=["bag_of_words", 'tf_idf', 'idf', 'avg_w2v', 'wt_w2v', 'brand_colour','cnn'], y=euclidean_distance,
                    mode='lines+markers',
                    name='lines+markers'))

fig.update_layout(title='Euclidean Distance Measurement of different components',
                   xaxis_title='Methods',
                   yaxis_title='Euclidean Distance')
fig.update_xaxes(
    ticktext=["Bag of Words", 'TF-IDF', 'IDF', 'Average Word2Vec', 'Weighted Word2Vec', 'Brand Colour','CNN'],
    tickvals=["bag_of_words", 'tf_idf', 'idf', 'avg_w2v', 'wt_w2v', 'brand_colour','cnn'],
)
fig.show()


fig = px.histogram( x=["Bag of Words", 'TF-IDF', 'IDF', 'Average Word2Vec', 'Weighted Word2Vec', 'Brand Colour','CNN'], y=euclidean_distance, histfunc='sum')
fig.update_xaxes(showline=True, linewidth=1, linecolor='black', mirror=True)
fig.update_yaxes(showline=True, linewidth=1, linecolor='black', mirror=True)

fig.update_layout(title='Euclidean Distance Measurement of different components',
                   xaxis_title='Methods',
                   yaxis_title='Euclidean Distance')
fig.show()


data.to_csv('Recommendations based on  Amazon Women Shirts data.csv',index=False)

	asin	brand	color	medium_image_url	product_type_name	title	formatted_price
0	B016I2TS4W	FNC7C	None	https://images-na.ssl-images-amazon.com/images...	SHIRT	Minions Como Superheroes Ironman Long Sleeve R...	None
1	B01N49AI08	FIG Clothing	None	https://images-na.ssl-images-amazon.com/images...	SHIRT	FIG Clothing Womens Izo Tunic	None
2	B01JDPCOHO	FIG Clothing	None	https://images-na.ssl-images-amazon.com/images...	SHIRT	FIG Clothing Womens Won Top	None
3	B01N19U5H5	Focal18	None	https://images-na.ssl-images-amazon.com/images...	SHIRT	Focal18 Sailor Collar Bubble Sleeve Blouse Shi...	None
4	B004GSI2OS	FeatherLite	Onyx Black/ Stone	https://images-na.ssl-images-amazon.com/images...	SHIRT	Featherlite Ladies' Long Sleeve Stain Resistan...	$26.26

	asin	brand	color	medium_image_url	product_type_name	title	formatted_price
0	B004GSI2OS	FeatherLite	Onyx Black/ Stone	https://images-na.ssl-images-amazon.com/images...	SHIRT	Featherlite Ladies' Long Sleeve Stain Resistan...	$26.26
1	B012YX2ZPI	HX-Kingdom Fashion T-shirts	White	https://images-na.ssl-images-amazon.com/images...	SHIRT	Women's Unique 100% Cotton T - Special Olympic...	$9.99
2	B001LOUGE4	Fitness Etc.	Black	https://images-na.ssl-images-amazon.com/images...	SHIRT	Ladies Cotton Tank 2x1 Ribbed Tank Top	$11.99
3	B003BSRPB0	FeatherLite	White	https://images-na.ssl-images-amazon.com/images...	SHIRT	FeatherLite Ladies' Moisture Free Mesh Sport S...	$20.54
4	B014ICEDNA	FNC7C	Purple	https://images-na.ssl-images-amazon.com/images...	SHIRT	Supernatural Chibis Sam Dean And Castiel Short...	$7.50
...	...	...	...	...	...	...	...
14333	B01FIVXRV4	DKNYC	Leopard	https://images-na.ssl-images-amazon.com/images...	SHIRT	DKNYC WOMENS LAYERED LOOK-BUILT IN FAUX SHIRT ...	$19.99
14334	B0753Z7N8W	susana monaco	Tanseite	https://images-na.ssl-images-amazon.com/images...	SHIRT	Susana Monaco Womens Cleo One-Shoulder Top, Xs...	$65.99
14335	B06Y1B4Z8L	Style&Co	Blue	https://images-na.ssl-images-amazon.com/images...	SHIRT	Blue 2 In 1 Tank Style Scoop Neck Cap Sleeve S...	$7.99
14336	B00WKXS1NI	Jerzees	Black	https://images-na.ssl-images-amazon.com/images...	SHIRT	Jerzees Ladies' 5.3 oz., 65/35 Easy-Care™ Polo...	$7.43
14337	B01FRGDGK2	Stella McCartney	Multi	https://images-na.ssl-images-amazon.com/images...	SHIRT	Stella McCartney Women's Striped Floral Tank	$139.97

The Mad Dresser : Data Science Demo Analysis Report¶

Contact Information¶

Objective¶

Data Acquisition¶